Evaluating the Performance of Weather Radars in South Africa: A Scientific Approach

Operating a weather radar in a developing country poses unique challenges. Severe rainfall events do occur from time to time over the interior of South Africa. Understanding these storms will improve our readiness, forecasting and prediction of these storms. In South Africa this is not always the case as budget constraints hampers the maintenance and training of these radars. Developing countries has also fallen behind in terms of new technologies and ways to measure radar rainfall estimates and new radar operation techniques as South Africa is still stuck with single polarized radars whereas weather radars in developed countries has dual polarized capabilities. Therefore new and unique observation techniques with the current resource that we have are needed to keep up with the developed countries. The interior of South Africa is an extremely important part of the country due to the agricultural activities and the amount of citizens living in highly dense townships often within floodplains. Therefore the need is high for highly accurate and reliable rainfall data within this area. The South African Weather Service (SAWS) has a large network of weather radars which includes 9 single polarized S band radars, 1 dual polarized S band weather radar, 2 mobile dual polarized X band radars and 5 C band radars.Three of these radars coverage area covers the interior of South Africa with one of them being a state of the art dual polarized weather radar. These radars often do not work to their full potential especially the dual polarized radar situated in Bethlehem, Gauteng. Availability of data for the entire network is also a problem with only 54% of data retrieved for 2014. Other radars that cover this area are the Irene weather radar and the Ottosdal weather radar. Developing new reliable and accurate techniques is very important and therefore the North West University has embarked on a project to develop our own rainfall infrastructure within the interior of South Africa. This has been done to assist the South African Weather Service in their effort to improve rainfall estimates. This infrastructure includes a C band weather radar, a high density rain gauge network consisting of 20 tipping bucket rain gauges, 5 Viasala multi-sensor weather transmitters and a Parsivel disdrometer which is the first time in South Africa that measurements using this type of disdrometer was done. The rain gauge network installed is also unique due to the fact that 20 rain gauges has been installed in the Mooi river catchment area which is 3294sq. km. This is the only high density network in South Africa compared to the network operated by SAWS of 300+ gauges over the entire South Africa. This paper is aimed at evaluating the performance of three radars operated by SAWS. The coverage area of all three these radars covers the rainfall observation infrastructure of the North West University. The data measured by the radar is compared to the measurement taken on the ground by the disdrometer. The correlation between these measurements gave a good indication on the performance of the radar. The Irene weather radar performed the best with a correlation of 66%, whereas the Ottosdal radar had a correlation of 59% and the Bethlehem radar 35%. This low correlation is because of challenges unique to South Africa which is not often recognized and understood by the international community. Rainfall estimates was also improved by calculating and implementing robust algorithm to stratify rainfall events and characterize them in mixed, convective and stratiform precipitation. Implementing a new unique Z-R relation to 3 different case studies showed an increase in accuracy compared to the Marshall and Palmer. For mixed precipitation the data improved from 66% to 87% compared to rain gauge data. For convective events correlation improved from 80% to 95% and for stratiform from 49% to 50%.